Exploring predictors of insomnia severity in shift workers using machine learning model.

Area of Science:

• Occupational Health
• Sleep Medicine
• Data Science

Background:

• Shift work disrupts circadian rhythms, leading to distinct insomnia features.
• Previous research has explored limited predictors of insomnia severity in shift workers.
• A data-driven approach is needed to identify key predictors and develop a robust insomnia prediction model for shift workers.

Purpose of the Study:

• To identify potential predictors of insomnia severity in shift workers using a machine learning (ML) approach.
• To evaluate the accuracy of an ML-based prediction model for insomnia in shift workers.
• To explore the unique predictors of insomnia severity in shift workers compared to non-shift workers.

Main Methods:

• Assessed predictors of insomnia severity in 4,572 shift workers and 2,093 non-shift workers.
• Utilized a general linear model with the least absolute shrinkage and selection operator (LASSO) for ML model development.
• Conducted additional analyses to assess interaction effects based on shift work schedules.

Main Results:

• Identified 41 key predictors from 281 variables, including demographic, physical health, job characteristics, and mental health factors.
• Shift workers showed stronger associations between insomnia severity and factors like work passiveness, authoritarian atmosphere, ease of waking, stress, and medication.
• The ML prediction model demonstrated good overall accuracy and specificity, with better F1 scores and recall for shift workers compared to non-shift workers.

Conclusions:

• The ML algorithm effectively identifies key predictors of insomnia severity in shift workers, incorporating workplace conditions.
• Findings align with traditional insomnia models but highlight unique shift work features.
• Developing comprehensive ML-based prediction models with identified key predictors is recommended for clinical applications and future research.